Metrics

Confidence Calibration: Aligning Conviction with Outcomes

Marcus Chen·Head of Analytics

September 15, 20247 min read

Professional traders often describe a "feel" for when a trade will work. But is this feeling reliable? Confidence calibration analysis answers this question with data.

What is Confidence Calibration?

Before each trade, you rate your confidence (1-10). Afterward, we compare these ratings with actual outcomes. Perfect calibration means:

High confidence trades win more often

Low confidence trades win less often

Your confidence accurately predicts probability

Common Calibration Problems

Overconfidence

High confidence ratings on trades that lose. This leads to:

Oversized positions on bad trades

Ignoring warning signs

Slower stop losses

Underconfidence

Low confidence ratings on trades that win. This leads to:

Undersized positions on good trades

Early exits before targets

Skipping valid setups

Random Confidence

No correlation between confidence and outcomes. This suggests:

Your intuition isn't trained

You need more deliberate practice

Focus on explicit criteria rather than feel

Improving Calibration

1. Define Confidence Criteria

Don't rate on vague "feel." Define what makes a trade high confidence:

Multiple confluent factors

Clear catalyst

Appropriate market conditions

Pattern recognition from past winners

2. Track and Review

Weekly, review trades by confidence level. Ask:

What did my high-confidence winners have in common?

Why did high-confidence trades lose?

What caused low confidence on eventual winners?

3. Adjust Sizing

Once calibrated, use confidence for position sizing:

8-10 confidence: Full size

5-7 confidence: 75% size

1-4 confidence: 50% size (or skip)

Using Practice—Process

The Confidence Calibration dashboard shows:

Your confidence vs. outcome correlation

Calibration curve (expected vs. actual win rate by confidence level)

Over/underconfidence by trade type

Improvement over time

Rate every trade honestly, and you'll develop a quantified sense of your own intuition's reliability.

Marcus Chen

Head of Analytics